Claude Fable 5: What It Costs and When to Use It

Claude Fable 5 went generally available on June 9, 2026, and per Anthropic's launch announcement it was the company's most capable generally available model, the first of a new Mythos-class tier that sits above Opus in the lineup rather than replacing it. The API identifier is claude-fable-5, and it ships with a 1M tokens (128k output) window and adaptive thinking always on: the model decides how hard to reason per request, with no extended-thinking toggle to manage.

Since September 1, 2026 the practical question has changed. Fable 5.1 sits in the same frontier slot at the same base price with cheaper cache reads, so the decision is whether a workload validated on Fable 5 is worth re-testing on 5.1, and when the frontier tier earns its price over the current Claude lineup, where Opus 5.5 is Anthropic's recommended starting model. Every current option is on the models hub. This guide sticks to verified facts.

Data verified September 23, 2026 against Anthropic's published pricing and docs. For scale: $10 in / $50 out per MTok was exactly double Opus 4.8's $5 in / $25 out per MTok at launch, and it is 2.5x Opus 5.5's $4 in / $20 out per MTok today, which is why the routing question in the last section matters more than the benchmark bragging rights.

Per Anthropic, Claude Fable 5 and Claude Mythos 5 share the same underlying model. Fable is the generally available version, shipped with additional safety measures around dual-use capabilities. Mythos 5 is invitation-only, reserved for approved organizations in Anthropic's defensive cybersecurity program. For most readers, Fable 5 was the one they could actually use; Fable 5.1 now fills that generally available slot.

No frontier model launch in recent memory has had a first month like this, and it is why so many people are searching for Fable 5 in the first place. The timeline:

1. 1

   June 9: launch

   Fable 5 goes generally available as Anthropic's most capable model and the first Mythos-class release.
2. 2

   June 12: US export controls applied

   Three days in, US export controls are applied to the model, restricting availability outside the United States.
3. 3

   June 30: controls lifted

   The export controls are lifted after just under three weeks.
4. 4

   July 1: redeployed globally

   Anthropic redeploys Fable 5 worldwide with new cybersecurity classifiers, detailed at anthropic.com/news/redeploying-fable-5.

At launch, Fable 5 held the best published SWE-bench scores of any generally available model, and Anthropic cited it as #1 on Cognition's FrontierCode benchmark. The verified numbers:

Data verified September 23, 2026. Note what these numbers do and do not say: SWE-bench measures resolving real GitHub issues, which maps well to agentic coding, but no single benchmark settles a model war. At the time, OpenAI's GPT-5.6 Sol led on terminal-agent benchmarks; that match-up is archived in Claude Fable 5 vs GPT-5.6 Sol. For the current generation, Anthropic's Opus 5.5 launch table puts Fable 5.1 at 55.8% on Terminal-Bench 4.0, and the cross-vendor view is in Claude Opus 5.5 vs GPT-6 Astra.

Claude Code v2.1.170 added Fable 5 behind the fable alias, and it was never the default model. Since v2.1.257 that alias selects Fable 5.1, so a workload that must stay on Fable 5 has to name it by its full model ID:

/model claude-fable-5

Two details from the launch. The alias best resolved to Fable 5 for organizations with access and to the latest Opus otherwise, so scripts using it degraded gracefully. And Fable 5 was unavailable under zero-data-retention agreements: if your org runs ZDR, check Anthropic's current terms before planning around any Fable model.

If you run agents in parallel, the selection question gets more interesting. On an AIDEN board, each feature card launches its own Claude Code session, so you can point Fable 5.1, or a pinned Fable 5, at the one gnarly cross-cutting feature while Opus 5.5 and Sonnet 5.5 handle the rest of the column. That per-feature routing is the whole economics of the frontier tier, more in Claude Code orchestration.

At Fable 5's launch, Opus 4.8 was half the price and Anthropic's recommended starting point for complex agentic coding. Both models have since been replaced, so read this as the launch-time comparison:

The gap was real: Fable 5 was measurably ahead on the hardest benchmarks, and on long autonomous runs where a wrong turn costs hours, the stronger model pays for itself. Reach for it when the work is genuinely hard: architecture-level refactors that cut across a codebase, multi-hour unattended runs, and drafting specs against a large repo where the 1M tokens (128k output) window earns its keep.

The same rule carries over to today's lineup, with a wider price gap: Opus 5.5 ($4 in / $20 out per MTok) is Anthropic's recommended starting model, and Fable 5.1 ($10 in / $50 out per MTok) is the escalation when higher Opus effort still falls short. The current decision rule is in the best Claude model for coding.